Preferred Hydrogen-Bonding Partners of Cysteine: Implications for

Joana ReisNicola ManzellaFernando CagideJeanne Mialet-PerezEugenio UriarteAngelo PariniFernanda BorgesClaudia Binda. Journal of Medicinal Chemistry ...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/JPCB

Preferred Hydrogen-Bonding Partners of Cysteine: Implications for Regulating Cys Functions Karine Mazmanian,†,‡,§ Karen Sargsyan,† Cédric Grauffel,† Todor Dudev,∥ and Carmay Lim*,†,⊥ †

Institute of Biomedical Sciences, Academia Sinica, Taipei 11529, Taiwan Chemical Biology and Molecular Biophysics Program, Taiwan International Graduate Program, Academia Sinica, Taipei 11529, Taiwan § Institute of Biochemical Sciences, National Taiwan University, Taipei 10617, Taiwan ∥ Faculty of Chemistry and Pharmacy, Sofia University, Sofia 1164, Bulgaria ⊥ Department of Chemistry, National Tsing Hua University, Hsinchu 300, Taiwan ‡

S Supporting Information *

ABSTRACT: The hydrogen-bonding interactions of cysteine, which can serve as a hydrogen-bond donor and/or acceptor, play a central role in cysteine’s diverse functional roles in proteins. They affect the balance between the neutral thiol (SH) or thiolate (S−) and the charge distribution in the rate-limiting transition state of a reaction. Despite their importance, no study has determined the preferred hydrogen-bonding partners of cysteine serving as a hydrogen-bond donor or acceptor. By computing the free energy for displacing a peptide backbone hydrogen-bonded to cysteine with amino acid side chains in various protein environments, we have evaluated how the strength of the hydrogen bond to the cysteine thiol/thiolate depends on its hydrogen-bonding partner and its local environment. The predicted hydrogen-bonding partners preferred by cysteine are consistent with the hydrogen-bonding interactions made by cysteines in 9138 nonredundant X-ray structures. Our results suggest a mechanism to regulate the reactivity of cysteines and a strategy to design drugs based on the hydrogen-bonding preference of cysteine.



INTRODUCTION Although cysteine (Cys) is the second least abundant amino acid (aa) residue (only ∼1.9% of all aa residues in proteins), it nevertheless plays diverse functional roles in proteins (see Chart 1).1 Cysteine can play a catalytic role by acting as (i) a potent nucleophile in enzymes such as cysteine proteases and ubiquitin ligases/deubiquitinases, (ii) an acid/base in abstracting or donating a proton to or from a substrate during the catalytic reaction, and (iii) a transition-state stabilizer (e.g., in stabilizing an oxyanion hole). It can also play a catalytic role by activating a water molecule, cofactor, or substrate or by participating in redox reactions catalyzed by enzymes in the thiol oxidoreductase family.2−4 Apart from a catalytic role, Cys can play a structural role to stabilize the protein by forming disulfide bridges5 or by binding metal ions such as Fe, Zn, Cd, and Hg to form mono/polynuclear clusters.6 In addition to its catalytic and structural roles, Cys can serve a regulatory role by acting as a switch to regulate the activity of proteins such as transcription factors, protein kinases, and metabolic enzymes.4,7 The wide variety of Cys functional roles can be attributed to the physicochemical properties and rich chemistry of its side chain.6 The Cys chemical reactivity in a protein at physiological pH is dictated by its solvent accessibility and its hydrogen© 2016 American Chemical Society

bonding interactions, which can tip the balance between the neutral thiol (SH) or thiolate (S−) to favor the protonated or deprotonated state8,9 and can stabilize charges in the ratelimiting transition state of a reaction.10,11 Among the 20 aa residues, Cys is found to be the least solvent-exposed residue in proteins.1 It can serve as a hydrogen bond (HB) donor when protonated as well as a HB acceptor in both protonated and deprotonated states.4,8−10,12 Thus, elucidating the factors governing the hydrogen-bonding interactions with the Cys thiol or thiolate is crucial for understanding and predicting its reactivity. Compared to studies on HBs to CH, NH, and OH, there are fewer studies on HBs to SH or S− despite their ubiquitous presence in proteins. Most of these studies have focused on HBs involving S formed by small molecules (e.g., hydrogen sulfide) to elucidate the directionality, nature, and strength of such HBs.13−18 The few studies on HBs to Cys in proteins have revealed the following: Received: August 11, 2016 Revised: September 16, 2016 Published: September 16, 2016 10288

DOI: 10.1021/acs.jpcb.6b08109 J. Phys. Chem. B 2016, 120, 10288−10296

Article

The Journal of Physical Chemistry B

computing the ΔGε for ε ranging from 4 to 80 according to eq 2:

Chart 1. Diagram Summarizing the Different Functional Roles of Cys in Proteins

ΔGε = ΔG1 + ΔΔGsolv ε

(2)

In eq 2, ΔG is the gas-phase free energy for eq 1, while ΔΔGsolvε is the solvation free energy difference between the products and reactants for a given ε. A negative ΔGε implies that Cys prefers to interact with the aa side chain X rather than the peptide backbone in an environment characterized by ε. The results were then compared with trends from a statistical survey of protein X-ray structures. They reveal that cysteines from protein X-ray structures usually form HBs in accord with the results from the computed free energies. The trends obtained help to understand the reactivity and diverse functional roles of cysteine. 1



METHODS Models Used. The aa residue side chains in their neutral or ionized form at physiological pH were modeled by the model compounds, as summarized in Table 1. All models were built using GaussView 5.23

Table 1. Compounds Modeling Amino Acid Side Chains and Peptide Backbone Groups

(1) Cys is found more often as a HB donor than as a HB acceptor in protein X-ray structures by a ratio of 5:1.19 (2) Compared to HBs to oxygen, those to the larger and more polarizable sulfur are longer19,20 with a mean H··· S(acceptor) distance of 2.80 ± 0.26 Å. (3) The Cys side chain in an α helix frequently forms HBs with the backbone oxygen, stabilizing the helix.20 (4) Intermolecular HBs formed between the Cys(S) and the backbone N can lower the pKa of a N-terminal Cys in a helix,21 whereas intramolecular HBs formed between the Cys(SH) and the backbone O can lower the pKa of active-site Cys.22 (5) HBs to a Zn-bound Cys(S−) in Zn finger proteins reduce the negative charge on all Zn-bound Cys(S−) atoms, even those with no hydrogen-bonding interactions.10 Although hydrogen-bonding interactions of the Cys side chain clearly depend on its partner, little attention has been paid to cysteine’s preference for its HB donor or acceptor partner. Hence, it is not known (i) if there is a preference among the aa residues or peptide backbone to accept/donate a HB from/to protonated Cys(SH) or deprotonated Cys(S−), (ii) which residues are preferred, and (iii) how the protein environment of the Cys would modulate this preference. To address this, we have focused on metal-free unmodified cysteines in this study and have examined the competition among the various groups in a protein to accept a HB from neutral Cys(SH) or donate a HB to protonated Cys(SH) or deprotonated Cys(S−). The outcome of this competition was assessed by the free energy, ΔGε, for displacing the peptide backbone (denoted as bkbn) that is hydrogen-bonded to Cys(SH) or Cys(S−) with an aa side chain X in a protein environment characterized by an effective dielectric constant ε; i.e., Cys:bkbn + X → Cys:X + bkbn

residue

model compound

chemical formula

backbone Arg+ Asn/Gln Asp−/Glu− Cys Cys− His0 His+ Lys+ Met Ser/Thr Trp Tyr

N-methylacetamide methylguanidinium acetamide acetate methanethiol methanethiolate imidazole imidazolium methylammonium dimethyl sulfide methanol 3-methylindole phenol

CH3CONHCH3 C2H8N3+ CH3CONH2 CH3COO− CH3SH CH3S− (CH)3N(NH) (CH)3(NH)2 CH3NH3+ (CH3)2S CH3OH C9H10N C6H6O

Calibrating the Geometry Optimization Method. To obtain reliable geometries of the hydrogen-bonded complexes involving sulfur, the experimentally determined structure of CH3SH24 was used to calibrate the geometry using M06 functionals (M06, M06-L, M06-HF, M06-2X)25 in combination with different Pople’s basis sets containing both polarization and diffuse functions. The four M06 functionals were chosen, as they account for the dispersion contribution and have been recommended for noncovalent interactions.26 The results in Supporting Table S1 show that the M06-2X functional with the 6-311++G(d,p) efficiently reproduced the experimentally observed CH3SH geometry. The computed C−S (1.82 Å) and S−H (1.34 Å) bond distances are identical to the measured ones, while the H−S−C and H−C−H angles are within 1.2° of the experimental values (96.5 and 109.8°, respectively). In previous studies, the M06-2X functional was found to reproduce the relative energies of Cys conformers computed using more accurate CBS-QB3, G3B3, G4MP2, and G4 composite methods.27 Out of 64 tested density functionals, M06-2X was found to be the most accurate one for predicting pKa values of aa residues and proton transfer between different residues.28,29 Thus, the M062X/6-311++G(d,p) method was used to compute the geometries and the vibrational frequencies

(1)

In eq 1, X is one of the possible hydrogen-bonding partners of Cys, viz., the side chains of Ser/Thr, Tyr, Asn/Gln, His, Arg, or Lys. The different protein environments of the Cys, which may be buried or solvent-exposed, were taken into account by 10289

DOI: 10.1021/acs.jpcb.6b08109 J. Phys. Chem. B 2016, 120, 10288−10296

Article

The Journal of Physical Chemistry B

Table 2. Root-Mean-Square Deviations (RMSDs) of the D···A and H···A Distances (Å) and the D−H···A Angles (deg) in the M062X/6-311++G(d,p) Optimized Structures from Those in the Corresponding CSD Structure CSD ID (HB type)

RMSD (Å) (D···A and H···A)

RMSD (deg) (D−H···A)

WUJWAZ (S−H···O) RONVEV (O−H···S; S−H···S) ADABIQ (N−H···S) YIRGEM (S−H···N) BABCOW (O−H···S; N−H···S)

0.05 0.04 0.05 0.10 0.11

0.4 5.5 0.9 1.6 4.8

Table 3. Comparison between Computed and Experimental Solvation Free Energies (in kcal/mol) and pKa Valuesa molecule AH

ΔGsolv80 (AH)

ΔGsolv80 (A)

pKa

acetic acid CH3COOH imidazolium (CH)3N(NH) phenol C6H6O methanethiol CH3SH methylammonium CH3NH3+ methylguanidinium C2H8N3+ methanol CH3OH N-methylacetamide CH3CONHCH3 acetamide CH3CONH2 dimethyl sulfide (CH3)2S 3-methylindole C9H10N

−6.7 (−6.7 ) −62.6 −6.4 (−6.6b) −1.0 (−1.2b) −73.4 −56.7 −5.0 (−5.1b) −9.7 (−10.0e) −9.6 (−9.7f) −1.4 (−1.5g) −5.9 (−5.9g)

−76.3 −9.9 (−10.2c) −70.1 −73.5 −4.4 (−4.6b) −11.1 (−11.2d) −96.0

4.9 (4.75) 7.2 (7.1) 10.2 (10.0) 10.7 (10.4) 10.9 (10.6) 13.6 (13.4) 15.8 (15.5)

b

a

Experimental values, where available, are in parentheses. The pKa values were computed using the experimental hydration free energy of the proton (−264.0 kcal/mol).35 bFrom Kelly, 2005.42 cFrom Lim et al., 1991.39 dFrom Vorobyov et al., 2008.43 eFrom Wolfenden, 1978.44 fFrom Wolfenden, 1981.45 gFrom Sitkoff et al., 1994.46

basis sets probed in previous studies28 and the ΔG1 of the proton (−6.28 kcal/mol35). The computed CH3SH deprotonation free energies in Supporting Table S2 show that three density functionals (B-971, B-98, and PBE1-PBE) along with the 6-31+G(2d,2p) basis set could reproduce the experimental ΔG1 for deprotonating CH3SH to within 0.1 kcal/mol. Since the B-98/6-31+G(2d,2p) method had been shown to perform well in computing the energies of hydrogen-bonding and weak noncovalent interactions,36,37 it was chosen to compute the ΔG1 of eq 1. Basis set superposition error was estimated using the counterpoise method and added to ΔG1 to yield the final energies reported herein. We also verified that the B-98/631+G(2d,2p) method yielded a very similar ΔG1 for eq 1 as the B-971 and PBE1-PBE functionals with the same basis set (see Supporting Table S3). Calibrating the Solvation Free Energy Calculations. The solvation free energy ΔGsolvε was estimated by solving Poisson’s equation using finite difference methods38,39 with the MEAD (Macroscopic Electrostatics with Atomic Detail) program,40 as described in previous works.41 The continuum dielectric calculations employed M062X/6-311++G(d,p) optimized geometries and natural bond orbital atomic charges as well as effective solute radii Reff (Supporting Table S4) that have been adjusted to reproduce the experimental solvation free energies and/or pKa of the compounds in Table 3. Database of Nonredundant Cys Structures. To verify the trends from the computed free energies, we retrieved 90°, and H···S−−AA or H···SH−bisector > 80°. Note that, if only the heavy atom−heavy atom distance (S···A < 4.3 Å or D···S < 4.1 Å) criterion was met but the other geometric requirements were not, a HB was not considered to be present in the X-ray structure.

− b

acetate (Asp /Glu ) imidazole (His0) dimethylsulfide (Met) methanol (Ser/Thr) methylthiol (Cys) acetamide (Asn/Gln) phenol (Tyr)



ΔG1

ΔG4

ΔG30

ΔG80

−11.8 −1.8 −1.1 −0.3 −0.2 0.5 0.7

−3.0 −1.7 −1.8 −0.1 −0.8 0.8 0.5

−0.6 −1.9 −2.0 0.2 −1.0 0.9 0.6

−0.3 −1.9 −2.0 0.2 −1.0 0.9 0.7

a ΔGε < −2.5 kcal/mol are highlighted in bold. bThe geometry of the CH3SH···CH3COO− hydrogen-bonded complex was constrainedoptimized at the M062X/6-311++G(d,p) level with the S···O− hydrogen-bonding distance fixed at 3.24 Å.

RESULTS Preferred HB Acceptors of Cysteine Thiol. The His imidazole N, Ser/Thr/Tyr hydroxyl O, the peptide backbone/ Asn/Gln carbonyl O, the Asp/Glu carboxylate O, and the Cys/ Met S can all accept a HB from protonated Cys0 thiol. To assess which of these aa residues prefer to accept HBs from the Cys H(S), their hydrogen-bonded complexes with methyl thiol were optimized at the M062X/6-311++G(d,p) level (see Supporting Table S5 for the HB parameters). During optimization of methyl thiol hydrogen-bonded to acetate, proton transfer occurred forming CH3S−···CH3COOH. To assess if such proton transfer might occur in the protein, the CH3SH···CH3COO− complex was optimized using the conductor-like polarizable continuum model (CPCM) in the Gaussian 0930 program with a dielectric constant ε ranging from 4 to 80. The results show that the proton remained on CH3SH with the sulfur 3.24 Å from the nearest carboxylate O. Hence, the CH3SH···CH3COO− hydrogen-bonded complex was reoptimized with the S···O− hydrogen-bonding distance constrained to 3.24 Å. The resulting structure and the other fully optimized geometries (Figure 2) were used to compute the free energies ΔGε (ε = 1−80) for replacing the peptide backbone with various HB acceptor side chains (Supporting Table S6). In the absence of the protein, the SH group strongly prefers to donate a HB to the Asp−/Glu− carboxylate (ΔG1 = −11.8 kcal/mol, Table 4), forming the shortest and most linear HB

compared with the other aa side chains (Supporting Table S5). The anionic Asp−/Glu− strongly polarizes the S−H bond: the NBO charge on S in the CH3SH···CH3COO− complex (−0.19e) is significantly more negative than that in free Cys0 (−0.04e). In contrast, the neutral aa side chains do not significantly polarize the S−H bond, as they increase the S negative charge in free Cys0 by only 0.01−0.04e (see Supporting Figure S2). In the presence of the protein, the preference for Asp−/Glu− is significantly dampened because the solvation free energy of the bulky anionic CH3SH···CH3COO− complex cannot compensate for the larger desolvation penalty of CH3COO−. Hence, the Cys thiol favors the Asp−/Glu− carboxylate over the backbone carbonyl group only when it is quite buried (ΔG4 = −3 kcal/mol). However, when Cys0 is partially/fully solvent exposed, it is not very discriminatory toward its HB acceptors: the ΔGε values differ by no more than 2 kcal/mol. Preferred HB Donors of Cysteine Thiol. In addition to serving as a HB donor, Cys0 can accept HBs from the backbone/His/Trp NH, Asn/Gln NH2, Ser/Thr/Tyr OH, the Cys SH, the Lys ammonium, and the Arg amino/imino groups. Interestingly, the HB parameters in Supporting Table S5 show that Cys0 formed shorter and more linear HBs with the 10291

DOI: 10.1021/acs.jpcb.6b08109 J. Phys. Chem. B 2016, 120, 10288−10296

Article

The Journal of Physical Chemistry B positively charged His+, Lys+, and Arg+ (H···S < 2.5 Å, N−H···S > 150°) than with neutral HB donors (H···S > 2.5 Å, N−H···S < 150°). Furthermore, Arg+ donates two hydrogen bonds to CysSH, unlike the other residues. The fully optimized structures of the hydrogen-bonded complexes with Cys0 as a HB acceptor in Figure 3 were used to compute the ΔGε free energies for replacing the peptide backbone with various HB donor side chains (Supporting Table S7).

Preferred HB Donors of Deprotonated Cysteine. Unlike Cys0, the Cys thiolate (Cys−) can only serve as a HB acceptor. In the fully optimized structures of Cys− hydrogenbonded to Lys+/His+, the thiolate became protonated (Figure 2b,c). Proton transfer from Lys+/His+ to Cys− was also observed in the CPCM-optimized hydrogen-bonded structures except when ε = 80. Hence, a Cys−···Lys+/His+ structure was obtained by constraining the S−···N distance to the value obtained using CPCM with ε = 80. The resulting Cys−···Lys+/ His+ structure and the other fully optimized geometries in Figure 4 were used to compute the ΔGε (Supporting Table S8).

Figure 3. Optimized hydrogen-bonded complexes of neutral Cys0 as a HB acceptor. The coloring scheme is the same as that in Figure 2.

The Cys0 thiol is more discriminatory as a HB acceptor than as a HB donor and prefers to interact with aa side chains rather than the peptide backbone (ΔGε all negative, Table 5).

Figure 4. Optimized hydrogen-bonded complexes of Cys− as HB acceptor. The coloring scheme is the same as that in Figure 2.

Table 5. Computed Free Energies (ΔGε in kcal/mol) for CysSH:bkbn + X → CysSH:X + bkbn, Where the Thiol Is a HB Acceptor, in a Medium with an Effective Dielectric Constant εa

Like the Cys0 thiol as a HB acceptor, Cys− is also discriminatory toward its HB donors, but the preference for charged HB donors depends on its relative solvent exposure: When the thiolate group is relatively buried, it strongly prefers positively charged aa side chains (Lys+, Arg+, and His+) as HB donors (negative ΔG4, Table 6). When it is partially solvent exposed, it prefers Lys+ and His+ only if it could accept a proton from these side chains forming Cys0···Lys0/His0 HBs (negative ΔG30, Table 6). On the other hand, when the thiolate group is fully solvent exposed, positively charged side chains become disfavored compared with the neutral peptide amide group because they form neutral Cys−···Lys+/His+/Arg+ complexes, whose solvation free energy gain cannot outweigh the cost of desolvating the cationic aa side chains (see Table 3). Unlike the Cys0 thiol, Cys− prefers the His0 imidazole among the neutral HB donors and disfavors the Ser/Thr side chain, as compared to the backbone amide, especially when buried (positive ΔG4, Table 6). When Cys is protonated, the size and flexibility of its HB donor seem to define HB complex formation: In comparison with the backbone amide group, the smaller and less rigid Ser/Thr hydroxyl group can better overlap with the sulfur lone pair, forming a stronger HB with Cys0, as evidenced by the shorter O···S (3.33 Å) and H···S (2.51 Å) distances and the larger O−H···S angle (143°) compared to those formed by the backbone amide (N···S = 3.47 Å, H···S = 2.78 Å, and N−H···S angle = 126°). However, when Cys is deprotonated, charge(S−)···dipole electrostatic

HB donor +

methylammonium (Lys ) imidazolium (His+) methylguanidinium (Arg+) methanol (Ser/Thr) acetamide (Asn/Gln) imidazole (His0) 3-methylindole (Trp) methanethiol (Cys) phenol (Tyr) a

ΔG1

ΔG4

ΔG30

ΔG80

−16.9 −13.6 −12.4 −3.5 −2.2 −1.9 −1.8 −1.6 −1.2

−7.3 −7.4 −7.7 −3.6 −2.3 −1.4 −1.6 −2.0 −1.2

−4.9 −5.6 −6.4 −3.6 −2.2 −1.3 −1.3 −2.0 −1.0

−4.6 −5.4 −6.3 −3.6 −2.2 −1.4 −1.3 −2.1 −1.0

ΔGε < −2.5 kcal/mol are highlighted in bold.

Whereas the Cys0 SH shows no strong preference for any HB acceptors except for Asp−/Glu− when buried, its S strongly prefers to accept a HB from positively charged aa side chains (His+, Lys+, or Arg+) and to a lesser extent from the Ser/Thr hydroxyl group, regardless of its solvent exposure. However, increasing solvent exposure of Cys0 attenuates the HB strength of the charged donors whose desolvation penalties exceed the solvation free energies of their hydrogen-bonded complexes. Nevertheless, Cys0 still prefers charged HB donors to the backbone amide even when it is solvent exposed. 10292

DOI: 10.1021/acs.jpcb.6b08109 J. Phys. Chem. B 2016, 120, 10288−10296

Article

The Journal of Physical Chemistry B Table 6. Computed Free Energies (ΔGε in kcal/mol) for CysS−:bkbn + X → CysS−:X + bkbn, Where the Cys0 Thiolate Is a HB Acceptor, in a Medium with an Effective Dielectric Constant εa ΔG1

HB donor +

methylammonium (Lys transfers proton to Cys−) imidazolium (His+ transfers proton to Cys−) methylammonium (Lys+)c methylguanidinium (Arg+) imidazolium (His+)c imidazole (His) phenol (Tyr) 3-methylindole (Trp) acetamide (Asn/Gln) methanol (Ser/Thr)

−122.4

ΔG4 b

−31.8

b

Table 7. Correlation between % Fraction of Side Chain X Winners ( f X,W) and ΔGε (in kcal/mol) for Cys:bkbn + X → Cys:X + bkbn in a Medium with an Effective Dielectric Constant ε

ΔG30

ΔG80

HB type

X

NX

f X,W

ΔG4

ΔG80

−3.4

−0.5

CysSH···S CysSH···O CysS−···H

Cys Ser/Thr Ser/Thr

708 103 103

59.4% 49.5% 46.4%

−0.8 −0.1 2.8

−1.0 0.2 1.1

b

b

−109.0b

−29.7b

−5.6b

−3.2b

−100.6c −90.1 −88.1c −3.6 −1.7 0.1 2.6 7.8

−19.7c −19.2 −17.1c −3.4 −1.4 1.6 2.7 2.8

4.3c 2.1 3.7c −3.0 −1.2 1.6 3.0 1.2

6.7c 4.4 5.8c −2.9 −1.2 1.6 3.0 1.1

rather than the peptide backbone, consistent with the small negative ΔGε (∼ −1 kcal/mol) for Cys in Table 4. The f X,W of ∼50% for Ser/Thr to accept a HB from protonated CysSH indicates that Cys0 has no preference to donate a HB to the Ser/Thr hydroxyl or the backbone carbonyl group, in accord with the near zero ΔGε for Ser/Thr in Table 4. In contrast, the f X,W for Ser/Thr to donate a HB to deprotonated Cys− is ∼46%, in line with the small positive ΔGε for Ser/Thr in Table 6.

ΔGε < −2.5 kcal/mol are highlighted in bold, whereas ΔGε > 2.5 kcal/mol are highlighted in italic. bBased on geometries optimized at the M062X/6-311++G(d,p) level in the gas phase where CysS− becomes protonated. cThe geometry of the hydrogen-bonded complex was constrained-optimized at the M062X/6-311++G(d,p) level with the CysS−···H(N) distance constrained to 3.06 and 3.00 Å for Lys+ and His+, respectively, to prevent proton transfer to CysS−. a



DISCUSSION The ability of Cys to form HBs has been challenged for a long time due to the low electronegativity of sulfur. Experimental and theoretical evidence15,50,51 now indicates that Cys can be a HB donor and/or acceptor and its hydrogen-bonding interactions influence its reactivity and thus biological functions.8,10,11,52 However, no study (to our knowledge) has determined the preferred hydrogen-bonding partners of Cys serving as a HB donor or acceptor. Here, we have elucidated how the strength of the HB to the Cys thiol/thiolate depends on its hydrogen-bonding partner and its local environment as well as the physical basis for the observed preference (see below). Physical Basis for the Selectivity of Cys HydrogenBonding Partners. The preferred hydrogen-bonding partner of Cys depends largely on the local protein environment (effective dielectric constant ε), which dictates the nature of the interaction stabilizing the HB. A buried Cys is quite selective toward its hydrogen-bonding partner, preferring charged aa side chains to neutral ones or the peptide backbone. For buried Cys residues, the nature of the electrostatic interactions dictates its preferred charged hydrogen-bonding partners: When the Cys is protonated, its weak S−H bond is polarized by HB interactions, inducing a dipole; hence, its HB strength depends on induceddipole(S−H)···charge/dipole interactions.17 On the other hand, when the Cys is deprotonated, its HB strength depends on charge(S−)···charge/dipole interactions.17 As induceddipole(S−H)/charge(S−)···charge interactions are more favorable than induced-dipole(S−H)/charge(S−)···dipole interactions, a buried Cys prefers negatively charged Asp−/Glu− as a HB acceptor (Table 4) and positively charged Lys+, His+, or Arg+ as HB donors (Tables 5 and 6). Solvent exposure of the Cys weakens its hydrogen-bonding partner selectivity. A solvent-exposed Cys is selective toward HB donors but not HB acceptors. Protonated Cys0 still favors positively charged Lys+, His+, or Arg+ as HB donors, but deprotonated Cys− favors neutral imidazole instead of positively charged aa side chains. Thus, the Cys− thiolate is stabilized by basic charged residues in a “dry” environment but by neutral aa residues when solvent exposed. Unlike buried Cys0, which favors Asp−/Glu− as a HB acceptor, solventexposed Cys0 exhibits no strong preference toward HB acceptors (|ΔG80| ≤ 2.0 kcal/mol, Table 4).

interactions become dominant in a low dielectric environment. As the dipole moment of the backbone amide or His0 imidazole (∼3.9 D) is roughly twice that of the Ser/Thr hydroxyl group (∼1.9 D), a buried Cys− prefers to accept a HB from the backbone amide or His0 imidazole rather than the Ser/Thr side chain. Preferred HB Partners of Cys from PDB Structures. To verify the preferred hydrogen-bonding partners of Cys predicted by the free energies ΔGε, we analyzed the hydrogen-bonding interactions made by Cys in 9138 nonredundant X-ray structures in the PDB containing 52,698 metal-free Cys residues (see Methods). Since eq 1 represents a competition between the backbone and aa side chains to form HBs with Cys, we first identified all competing backbone and side chain donor/acceptor heavy atoms within 4.3 Å from the Cys S. Given the Cys protonation state assigned by the REDUCE program, each potential side chain X listed in Tables 4−6 was evaluated to see if it formed a HB with Cys according to geometric criteria (see Methods) and, if it did, whether its rival backbone acceptor/donor formed a HB or not. In the competition between side chain and backbone acceptor/donor to form HBs with Cys, let NX,W denote the number of side chain X winners (i.e., the number of side chain X that formed a HB with Cys but not the competing backbone) and let NX,L denote the number of side chain X losers (i.e., the number of side chain X that did not form a HB with Cys but the backbone did). Since hydrogen atoms are not seen in the X-ray structures and were added using the REDUCE program (see Methods), we attempted to reduce errors in the PDB analysis by using sufficient samples for statistical analyses. Thus, we computed the % fraction of side chain X winners (f X,W) only when the total number of X competitors, NX = NX,W + NX,L, exceeded 100. NX was found to be >100 only for the side chains listed in Table 7, but it was zero or